EDA – Week 5
Comparing Batches
I recently finished grading your “comparing batches” homework and here are some remarks on some common issues where you lost points.
Issue 1: What is the Point of the Assignment?
We want to compare batches in a simple way. A simple comparison is say something like “Batch 2 tends to be 10 smaller than Batch 1”. We can make a simple comparison when the batches have similar spreads. In both of your problems, this wasn’t the case. The reason why we did all of the work (construct a spread vs level plot, etc) was to find a reexpression so that the is not a dependence between spread and level and the batches have similar spreads on the new reexpressed scale. If you have found a reasonable reexpression, then please compare the batches on the new scale. It is not okay to say that one batch is larger than another batch — we’d like to know how much larger. Many of you didn’t do this on your homework.
Issue 2: Did My Reexpression Work?
The reason why we reexpress is to remove the dependence between spread and level. How can one check if the reexpression works? Well, just construct a new spread-level plot on the reexpressed data and comment if there remains a dependence between spread and level. If there is some dependence, suggest the choice of a new reexpression that may help.
Issue 3: Be Nice to Your Grader
I am grading your homework and I really appreciate if you give me what I am calling “tidy work”. Here are some characteristics of tidy work:
- Just display the relevant R work. Don’t show me anything that is not relevant to what is being asked.
- Always explain what each graph is showing us. I don’t like naked graphs that have no accompanying explanation.
- Explain, explain, explain — I will never take points off if you write too much.
Last, please contact me if you don’t understand the question or if you have any R issues. It is not acceptable to simply say “this function didn’t work” on your homework. Generally, I try to be quick in responding to student requests.
Your Assignment this Week
Wait … it appears that you don’t have any work due this week. Enjoy the cooler weather and we will be looking at reexpressions more thoroughly next week.
EDA Trivia Question — send answer to albert@bgsu.edu
One of John Tukey’s collaborators was a famous British statistician who taught at Yale. (Hint — this statistician is famous for his “quartet”.) . This collaborator wrote a book on statistical computation using a particular programming language. What was the name of the statistician and name of that programming language?
Comments are closed.